Tidy code (4851e365) · Commits · Chaos / chaos.juicer

PBS/scripts/split_rmdups.awk

+13 −11

Changes for PBS/scripts/split_rmdups.awk: 13 added lines, 11 removed lines.
Original line number	Diff line number	Diff line
		@@ -35,13 +35,14 @@ BEGIN{
		if (tot >= 1000000) {
		if (p1 != $1 \|\| p2 != $2 \|\| p4 != $4 \|\| p5 != $5 \|\| p8 != $8) {
		print "this is the beginging part of split rmdups awk";
		cmd=sprintf("qstat \| grep osplit%s \|cut -c 1-7", groupname);
		cmd=sprintf("qstat \| grep osplit%s \|cut -c 1-8", groupname);
		cmd \|& getline jID_osplit;
		sname=sprintf("%s_msplit%04d_", groupname, name);
		sysstring1=sprintf("qsub -l %s -o %s_%s.log -j oe -q %s -N DDuP%s%s -l mem=4gb -l nodes=1:ppn=1:AMD -W depend=afterok:%s <<-EOF\nawk -f %s/scripts/dups.awk -v name=%s/%s %s/split%04d;\nEOF\n", walltime, outfile, sname, queue, name, groupname, jID_osplit ,juicedir, dir, sname, dir, name, dir, name);
		sysstring1=sprintf("qsub -l %s -o %s_%s.log -j oe -W group_list=cu_10027 -A cu_10027 -q %s -N DDuP%s%s -l mem=4gb -l nodes=1:ppn=1:thinnode -W depend=afterok:%s <<-EOF\nawk -f %s/scripts/dups.awk -v name=%s/%s %s/split%04d;\nEOF\n", walltime, outfile, sname, queue, name, groupname, jID_osplit ,juicedir, dir, sname, dir, name, dir, name);
		system(sysstring1);
		cmd1=sprintf("qstat \| grep DDuP%s%s \| cut -d ' ' -f 1",name,groupname);
		cmd1=sprintf("qstat \| grep DDuP%s%s \| cut -d ' ' -f 1 \| cut -d '.' -f 1-2",name,groupname);
		cmd1 \|& getline jID;

		close(cmd1);
		if (name==0){
		waitstring=sprintf("%s",jID );
		@@ -65,9 +66,11 @@ END {
		sname=sprintf("%s_msplit%04d_", groupname, name);
		print "This is the begining of END step, submitting _dedup job in END step";
		sysstring2=sprintf("qsub -l %s -o %s_%s_dedup%s.log -j oe -q %s -N DDuP%s%s -W depend=afterok:%s <<-EOF\nawk -f %s/scripts/dups.awk -v name=%s/%s %s/split%04d;\nEOF\n", walltime, outfile, sname, name, queue, name, groupname, jID_osplit ,juicedir, dir, sname, dir, name, dir, name);
		system(sysstring2);
		sname=sprintf("%s_msplit%04d_", groupname, name);
		system(sysstring2);

		cmd2=sprintf("qstat \| grep DDuP%s%s \| cut -d ' ' -f 1",name,groupname);
		cmd2=sprintf("qstat \| grep DDuP%s%s \| cut -d ' ' -f 1 \| cut -d '.' -f 1-2",name,groupname);
		cmd2 \|& getline jID;
		close(cmd2);
		if ( name==0 ){
		@@ -83,23 +86,22 @@ END {
		close(cmd3)
		print "below is the waitingstring1 and waitingstring2";
		print waitstring;
		sysstring = sprintf("qsub -l %s -o %s_catsplit.log -j oe -q %s -N CtSplt%s -W depend=afterok:%s <<-EOF\ncat %s/%s_msplit_optdups.txt > %s/opt_dups.txt; cat %s/%s_msplit_dups.txt > %s/dups.txt; cat %s/%s_msplit*_merged_nodups.txt > %s/merged_nodups.txt; \nEOF\n", walltime, outfile, queue, groupname, waitstring, dir, groupname, dir, dir, groupname, dir, dir, groupname, dir, dir);
		print waitstring2;
		print "below is te waited job status before qsub"
		print $waitedjobstatus;
		print "below is the waited job status before qsub"
		print waitedjobstatus;
		print "submitting _catsplit jobs";
		sysstring = sprintf("qsub -l %s -o %s_catsplit.log -j oe -q %s -N CtSplt%s -W depend=afterok:%s <<-EOF\ncat %s/%s_msplit_optdups.txt > %s/opt_dups.txt; cat %s/%s_msplit_dups.txt > %s/dups.txt; cat %s/%s_msplit*_merged_nodups.txt > %s/merged_nodups.txt; \nEOF\n", walltime, outfile, queue, groupname, waitstring, dir, groupname, dir, dir, groupname, dir, dir, groupname, dir, dir);
		system(sysstring);
		cmd4=sprintf("qstat \| grep CtSplt%s \| cut -d ' ' -f 1",groupname);
		cmd4=sprintf("qstat \| grep CtSplt%s \| cut -d ' ' -f 1 \| cut -d '.' -f 1-2",groupname);
		cmd4 \|& getline jID_catsplit;
		close(cmd4);
		print "below is the _catsplit id";
		print jID_catsplit;
		print "submitting _rmsplit jobs";
		sysstring=sprintf("qsub -l %s -o %s_rmsplit.log -j oe -q %s -N RmSplt%s -W depend=afterok:%s <<-EOF\n rm %s/_msplit_optdups.txt; rm %s/_msplit_dups.txt; rm %s/_msplit_merged_nodups.txt; rm %s/split*;\nEOF", walltime, outfile, queue, groupname,jID_catsplit, dir, dir, dir, dir);
		print "submitting _rmsplit jobs";
		system(sysstring);
		cmd=sprintf("qstat \| grep RmSplt%s \| cut -d ' ' -f 1",groupname);
		cmd=sprintf("qstat \| grep RmSplt%s \| cut -d ' ' -f 1 \| cut -d '.' -f 1-2",groupname);
		cmd \|& getline jID_rmsplit;
		print "below is the jID_rmsplit:";
		print jID_rmsplit;
		}

Admin message